Hacker News Reader

Hacker News

Comment by theandrewbailey | Hacker News Reader

Full thread

theandrewbailey

On the first graph, that last label makes me distrust the entire article.

View on HN

Possibly related:

02 Jan 2021 is a part of 2020 week 53.

I've noticed a few labeling issues caused by this calendar quirk.

Yep, that's due to someone calling

    strftime(buf, buflen, "%b %e, %G", &tm);

when they should be calling

    strftime(buf, buflen, "%b %e, %Y", &tm);

In Java it's worse, for SimpleDateFormat (the most often used date formatter in Java) YYYY (uppercase) is "week year" (strftime %G), the correct way to get the year (strftime %Y) is yyyy (lowercase). It's very easy to get it wrong; the correct way to get the full ISO timestamp is "yyyy-MM-dd'T'HH:mm:ss.SSSXXX".

It kind of makes sense. People are going to think of 2020 and 2021 as one long year. Maybe 2022, too.

It's an ISO 8601 vs Gregorian calendar peculiarity, not inaccuracy. See the date section here https://en.wikipedia.org/wiki/ISO_8601, another example of a week 53.

What about it? The fact that it says January 2 while the caption says it's through January 7? I'm assuming that's just because the labels are generated at 11 day increments and Jan 7 isn't 11 days away from the 2nd.

It says 2020 instead of 2021.

Ah, that's definitely strange. A bug in the graphing software, maybe?

I think it's more likely that there were one or more manual steps in generating the graph, and someone just kept going with 2020 instead of typing 2021. The author should explain what went wrong there if this reaches their attention.

Or it's a typo, and who cares?

Maybe a date formatting problem? I saw something similar in the past due to passing the wrong format string, and Jan 2 was the Saturday of the "last week" of 2020, before the "first week" of 2021?

Can you clarify which graph and which label? The first graph I see is the number of new cases in Ireland over the past year, and I don't see any labels that seem suspicious.

The final label on the x-axis says January 2, 2020 instead of 2021. I assume the GP was being partly facetious

What "last label" are you referring to?

https://www.google.com/search?q=ireland+covid+cases

Same graph. Happy?

We really need to get away from this meme of "I found a typo, therefore I'm going to discard the article". In theory it's good epistemic hygiene, but in practice it almost always cashes out to, "I'm going to engage in extremely selective sophistry so I can ignore information that doesn't agree with my existing worldview."

Reply on news.ycombinator.com